PROJECT PROPOSAL RESPONSE OF THE GREAT PLAINS NOCTURNAL LOW-LEVEL JET (NLLJ) TO VARIOUS WIND ENERGY SCENARIOS AS SIMULATED IN WRF.

Transcription

1 PROJECT PROPOSAL RESPONSE OF THE GREAT PLAINS NOCTURNAL LOW-LEVEL JET (NLLJ) TO VARIOUS WIND ENERGY SCENARIOS AS SIMULATED IN WRF Daniel Gilbert Department of Geography and Earth Sciences University of North Carolina at Charlotte Charlotte, NC May 26, 2012

2 Introduction The utilization of wind energy is rapidly growing in popularity as a solution to the world s need for sustainable energy. The American Wind Energy Association (AWEA) is currently aiming for twenty percent of US power to come from wind energy by the year 2030 (Salerno et al. 2008). While wind energy offers a number of significant benefits such as water conservation, energy security, and reduction in greenhouse gases it does have the potential to change patterns in local meteorology. The plausibility of these impacts is evident when considering that the total kinetic energy dissipation in the atmosphere is less than 1 PW and the human demand for energy is projected to surpass 20 TW by 2100 (Keith et al. 2004). Research indicates that the placement of wind energy turbines may have implications for regional weather patterns through changes in wind speed and direction, low-level moisture, temperature, and turbulence (Baidya Roy et al. 2004; Baidya Roy 2010; Lu 2011). This research will, through the use of the Weather Research and Forecasting (WRF) model, investigate the impacts wind energy turbines may have on the Great Plains southerly nocturnal low-level jet (NLLJ). One of the largest North American land based wind energy resources is located in the U.S. Great Plains (Figure 1). The NLLJ, a low-level wind maximum that tends to occur within the nocturnal boundary layer, is largely responsible for this significant wind resource. The enhanced magnitude of the NLLJ in this region is primarily caused by diurnal fluctuations of baroclinicity associated with sloping terrain and a decoupling effect known as the inertial oscillation, which is induced by a reduction in vertical mixing. This study will investigate the following potential feedbacks of wind energy on the NLLJ. 1) The two aforementioned processes will be analyzed to determine how they may change in response to wind farms and how those changes will feedback to NLLJ formation. It is 1

3 expected that wind farm induced turbulence will create a well-mixed planetary boundary layer (PBL) and decrease static stability resulting in a decrease of the inertial oscillation. Increased mixing may also act to decrease baroclinicity and result in weaker NLLJs. 2) It is uncertain how wind farm size, capacity density and location will play a role in the nature of NLLJ response. Various combinations of installed capacity and capacity density will be explored to determine how siting may provide a source of mitigation. Individual case studies will be used to determine how sensitive NLLJ changes are to the jet s location relative to the wind-farm. It is expected that the largest, highest density wind farms will have the most significant impact. The Great Plains LLJ contributes significantly to the large amounts of seasonal rainfall that are observed throughout late spring in the central plains (Helfand and Schubert 1995; Higgins et al. 1996). It is estimated to transport more than one third of the total water vapor into the central plains during the warm season. A decrease in the large amount of moisture advection that the LLJ provides would significantly affect agriculture in this region. With the many nonlinearities that exist in atmospheric sciences, it is not unreasonable to assume that there may be unforeseen negative consequences from a large change in the LLJ. Previous Research Explaining and Defining The Great Plains Nocturnal Low Level Jet The importance of low-level jet (LLJ) streams and the mechanisms responsible for their formation have been the subject of much interest in the field of meteorology and more recently in the development and implementation of wind energy. The significance of the Great Plains LLJ to severe weather has been established for some time (Fawbush et al. 1951). The vast majority of research describing the influence of the LLJ on severe weather has been focused primarily on the large amounts of moisture that the Great Plains LLJ transports from the Gulf of Mexico into the 2

4 Central United States (Strensrud 1996; Higgins et al. 1997; Tollerud et al. 2008). Previous research has indicated that the interaction between a LLJ and an upper-tropospheric wave creates an atmospheric setting that is capable of generating deep convection as the LLJ provides a source of low-level moisture convergence and the upper-tropospheric wave provides an additional source of lift (Uccellini 1980, 1979). Wang and Chen (2009) utilized over 20 years of reanalysis data to explain the link between the LLJ and the springtime rainfall maximum that occurs over the Great Plains. It was found that from May to June the southerly flow of the LLJ is further enhanced by the propagation of baroclinic waves from west to east. The LLJ and these waves interact to enhance convergence and convection. Moving into July, the synoptic pattern becomes increasingly dominated by an upper-level ridge classified by lower values of upper-level divergence and low-level moisture convergence. Coupling with an upper-tropospheric wave is not a requirement for destabilization, as research has indicated that a nocturnal LLJ promotes convection solely by providing a source of low-level convergence (Blackadar 1957; Modahl 1979). In addition to a source of low-level convergence, a LLJ provides a significant amount of low-level shear necessary to change the mode of convection (Blackadar 1957; Rasmussen 1998). Observations of nocturnal convection have indicated that the presence of a LLJ shifts the convective mode toward quasilinear convection with possible embedded supercells. It was also found that a LLJ can act to increase the intensity of convection through dynamic pressure forcing (Kis and Straka 2009). It is clear that a reduction in the magnitude of the LLJ has the potential to impact precipitation patterns over the central US. There has been a considerable amount of variability in the way that researchers have defined a LLJ. The American Meteorological Society defines a low-level jet as being a jet stream that is typically found in the lowest 2 to 3 km of the troposphere. While this definition 3

5 may sound concise, it leaves many characteristics of a low-level jet open for interpretation such as; the vertical distribution of velocity that must be present, the magnitude of maximum wind speed that must be achieved in order to classify a low-level maximum as a LLJ, and the horizontal spatial limitations of a LLJ (i.e. where does the LLJ start and where does it end?). The first descriptions of a low-level nocturnal wind maximum came from observations in Africa as early as 1938 (Fraquharson 1939; Goualt 1938; Wagner 1939). The term low-level jet was first introduced by Means (Means 1952, 1954) in his description of the strong nocturnal LLJ that develops in the Great Plains of the United States. Research on the Great Plains nocturnal LLJ has been mainly centered about four general categories; the mechanisms responsible for LLJ formation and the frequent location of strong LLJs over the Great Plains, the climatology of the LLJ, the large amounts of moisture advection that the LLJ provides to the Central United States, and the contribution of the LLJ to deep convection in Great Plains. Research between these four categories has typically lead to varying definitions of a LLJ. Early research lacked a concise definition for a LLJ. Blackadar (1957) was responsible for explaining the nocturnal occurrence of the LLJ as being caused by the decrease in eddy viscosity that occurs in a stable boundary layer. In the author s research describing the connection between viscosity and the development of a LLJ it was accepted that an increase in horizontal velocity with increasing height was classifiable as a LLJ. The concept of the inertial oscillation was first introduced here as the author observed the clockwise rotation of the winds with respect to time. The supergeostrophic wind speeds coupled with this oscillation provided ample evidence that the formation of a nocturnal LLJ must be related to a decrease in eddy viscosity that is characteristic of a nocturnal boundary layer. 4

6 Holton (1967) realized that the amplitude and shape of the inertial oscillation could not explain why the strongest nocturnal LLJ events tend to occur over the sloping terrain of the US Great Plains. The author s research established diurnal variations in baroclinicity as being a significant forcing mechanism responsible for generating the wind speeds observed within the LLJ. By creating a mathematical model approximately representative of boundary layer flow, he was able to show that the diurnal temperature cycle over sloping terrain forces diurnal oscillation of the wind by providing a source of gravitational potential energy. Holding eddy viscosity and heat diffusion constant, a hodograph was generated for a neutrally stable boundary layer and for an isothermal boundary layer. The neutrally stable boundary layer contained a much larger oscillation supporting the temperature field as being a forcing mechanism. This mechanism and the Blackadar (Blackadar 1957; Buajitti 1957) mechanism have both received further validation from increasingly complex mathematical models (Shapiro and Fedorovich 2009; Shapiro and Fedorovich 2010) and are the two most frequently cited mechanisms responsible for the Great Plains nocturnal LLJ. According to a definition provided by Stull (1988), the LLJ is frequently located 100 to 300 m above ground level (AGL) and occasionally reaches a height 900 m AGL. The definition given by Stull (1988) appears to limit the vertical limits of the LLJ to the planetary boundary layer (PBL). Requiring a wind maximum to occur within the PBL in order to meet classification as a LLJ is a logical restriction when considering that the nocturnal acceleration of the wind occurs as a result of boundary layer processes (i.e. diurnal oscillation in frictional effects). Specific elevation ranges have been given throughout research with a general consensus that the maximum height at which a LLJ typically occurs is 1000 meters AGL (Werth et al. 2011; Jiang et al. 2007; Storm and Basu 2010). 5

7 Kraus et al. (1985) described various mechanisms that have been observed to induce the onset of a LLJ, including many that are not the result of a nocturnal PBL. These mechanisms include; synoptic-scale baroclinicity, fronts, advective accelerations, flow patterns associated with mountainous terrain, and land and sea breezes. Limiting classification of a LLJ to those maxima that are associated with large horizontal and vertical shear was proposed by Reiter (1963, 1969). The author argued that wind maxima occurring within a stable boundary, but with low horizontal shear should be named inversion wind maxima in order to remain consistent with the definition of a jet stream. Similarly, Chen et al. (1994) argued for the label of boundary layer jet (BLJ) to be reserved for those wind maxima associated with diurnal fluctuations in PBL mixing and the term LLJ to be reserved to those low-level wind maxima associated with synoptic forcing. Rife et al. (2010) also made a clear distinction by designating low-level maxima occurring as a result of PBL processes as a nocturnal low-level jet (NLLJ) and designating maxima from synoptic forcing as a low-level jet (LLJ). As nocturnal stable PBL processes are considered to be primarily responsible for enhanced LLJ formation in the Great Plains, the research conducted herein will focus on the NLLJ. In previous climatology studies of the Great Plains LLJ (Bonner 1968; Song et al. 2005; Whiteman 1997) the horizontal wind shear has been neglected and the system of classification employed has been based solely on the vertical profile of the horizontal wind. This classification system is partly a result of limitations in data, with most data utilized in climatology being from wind profilers and sondes. The classification schemes used in these climatology studies required a large amount of vertical shear, especially between the level where the maximum occurred and the level just above the maximum. Bonner (1968) specified a maximum LLJ height of 3 km and required that a fifty percent decrease in wind speed occur at some point between height of the 6

8 maximum and 3 km. Later climatologies required that the decrease occur before 2 km (Song et al. 2005; Whiteman 1997). The required decrease in horizontal wind speed above the height of the LLJ maximum prior to PBL top will be utilized in the definition used in this research. The specific set of criteria to be used in classifying a NLLJ will be discussed in further detail in the Data and Methods section. Wind Energy Mitigating climate change by transitioning to a greater reliance on alternative sources of energy has been the subject of much debate. The Intergovernmental Panel on Climate Change (IPCC) has supported the position that we currently have the energy technologies that are needed to stabilize carbon emissions at less than double the preindustrial concentration over the next hundred years. There is also a portion of the scientific community that believes current energy technologies are insufficient to stabilize carbon emissions and that we must focus research efforts on more revolutionary technologies that are currently far from implementation (Hoffert et al. 2002; Pielke 2010). Pacala and Socolow (2004) presented an assessment of how currently employed alternative energy technologies can be expanded to stabilize CO 2 concentration at 500 parts per million (ppm). In order to stabilize concentrations at this level the authors proposed that the current emission rate of 7 billion tons of carbon per year (7 GtC/year) must be maintained despite current projections placing 2054 emissions around 14 GtC/year. It is important to note that continued stabilization past 2054 would require a decrease in emission rate from 7 GtC/year. Fifteen different technologies currently available for decarbonization such as; increased energy efficiency, decarbonization of electricity and fuels (alternative energy), and biological storage were reviewed with seven of them being theoretically expanded to stabilize emissions. These 7

9 technologies were referred to as stabilization wedges. In order for wind energy to fit the bill of being one of these wedges, worldwide installed capacity would need to be increased to 2000 GW. The installed wind capacity in 2004 was 40 GW at peak capacity meaning capacity would be required to increase by 50 times in order to meet this goal. When expanding and developing an energy source it is essential to have an assessment of the potential of that energy source. Before continuing, there are two terms that must be defined; capacity factor (CF) and capacity density (CD). Capacity factor is the ratio of actual power production to the power that would be produced at maximum rated output. It provides a measure of a wind farms actual production, which is always much lower than its nameplate/maximum rated capacity. Capacity density is the ratio of a wind farm s installed capacity to the area encompassing the wind farm. Archer and Jacobson (2005) quantified global wind power potential by utilizing five years of surface and sounding observations and performing a statistical analysis on one of those years to determine wind resources at hub height (80 m). This was accomplished by considering placement of six 1.5 MW turbines per square kilometer (CD of 9 W/m 2 ) over the portion of the earth s land area meeting class 3 resource criteria or greater ( 6.8 m/s). From this method it was estimated that the land-based global wind power potential is approximately 72 TW. With global energy demand in 2001 being between 1.6 and 1.8 TW, this estimate indicates that wind energy resources surpass demand by a considerable amount. Other research has suggested that the atmospheric feedbacks of large-scale wind energy development place limitations on global wind power potential (Adams and Keith 2011; Keith et al. 2004). Adams and Keith (2011) addressed this issue by simulating various wind farm size and density combinations in the Advanced Research Weather Research and Forecast model 8

10 (WRF-ARW) (Shamarock et al. 2008). It was shown that as capacity density increases the winds continue to slow down and capacity factor decreases. Archer and Jacobson (2005) assumed a constant capacity factor of 48% for their capacity density of 9 W/m 2, which is much higher than values commonly observed. It was found that beyond a capacity density of a few W/m 2, additional turbines do not increase energy production. These results indicate that previous estimates of wind energy resources are significantly overstated. Research herein will determine if this maximum threshold of power production also carries over to NLLJ response. Wind Energy and Its Impact on the Atmosphere Keith et al. (2004) looked directly at the climate impacts of large-scale wind power by using two general circulation models. The authors found that introducing perturbations to the exchange coefficients for momentum, heat, and moisture as well as making changes to the drag coefficient by modifying roughness length would act to provide an accurate simulation of the interaction between wind turbines and the atmosphere. An explicit drag scheme was also introduced for the purpose of simulating varying energy extraction rates with varying wind velocity. The increases in turbulent transport that wind turbines induce were neglected. Keith et al. (2004) found that the change in the mean global temperature for the 20 years they were modeling was virtually negligible while the response of peak-seasonal temperatures exceeded ± 2 C. Most interesting was the response within the northern extratropics where the magnitude of response was just as significant outside of the wind farms as it was within. The sign of atmospheric response differed between regions and the patterns of the response were similar for both models and all drag perturbations indicating that changes in climatic response are not the result of local surface energy budgets. A key point made in this research is that when estimating the total amount of wind resources available in a region, it is not at all accurate to assume that all 9

11 of the wind is available for extraction of power because dynamics impose a limitation on the amount of power that can be extracted. To better explain the results of Keith et al. (2004), Kirk-Davidoff and Keith (2008) utilized two models; a simple shallow-water model and a general circulation model in aquaplanet mode. The shallow-water model was run with a damping term introduced into the shallow-water vorticity equation to represent a surface roughness anomaly. Anomalies were introduced in the GCM by increasing surface roughness length. Several cases were run with varying patterns of zonal flow. It was found that for the case of a latitudinally varying Coriolis parameter the surface roughness anomaly generated a stationary Rossby wave, partially explaining the observed patterns of climate response in Keith et al. (2004). The GCM produced a similar, yet more complex response. Barrie and Kirk-Davidoff (2010) also used a general circulation model to simulate the impact of wind turbines on the atmosphere. Wind turbines were represented by increasing surface roughness. The model simulated a wind farm covering 23 percent of North America over a period of six years. The authors described the anomalies in their simulation as being similar to those observed by Kirk-Davidoff and Keith (2008). One important aspect of this research was that it more closely examined the downstream impacts of wind farms. This was done by comparing the wind farm simulation with the control simulation to observe how far reaching the anomalies from the wind farm extended. The results showed that initial perturbations in zonal wind occurred over the wind farm, as would be expected, then spread out and caused rather large perturbations as the model continued. The effects of the wind farm were most prominent in the Pacific after one week. Further analysis was not possible beyond two 10

12 weeks because the model run containing the wind farm had diverged significantly from the control run. Baidya Roy et al. (2004) utilized a mesoscale model to simulate the impacts of a wind farm over the Great Plains region of the United States. The wind turbines were represented as elevated sinks of resolved kinetic energy and sources of turbulent kinetic energy (TKE). This method was employed because the higher spatial resolution of a mesoscale model provides for multiple vertical levels of the model to intersect the turbine. However, the horizontal spatial resolution in a mesoscale model is still far too low to resolve the flow patterns induced by wind turbines. The kinetic energy was removed by use of a constant power coefficient. This is not an accurate representation of the power coefficient, which varies non-linearly with wind speed. Three simulations were conducted; a control run without wind turbines, a run in which the wind turbines extracted energy from the atmosphere, and a run in which wind turbines extracted energy and increased turbulent kinetic energy (TKE). Changes in thermodynamic variables did not occur in the run in which the turbines only extracted energy, but large changes were evident in the run representing the wind turbines as a source of energy sink and a TKE source. The authors observed a mixing down of upper level air and a mixing up of lower level air, which was observed to be responsible for observed warmer nocturnal surface temperatures. They considered this effect to be negligible during the daytime because the atmosphere is already wellmixed. In the simulation in which only energy was extracted, the authors observed a decrease in the wind speed within 500 meters around the hub height. They observed a decrease in the magnitude of the flow in a vertical layer reaching an altitude greater than 1 km for the scenario in which the wind farm extracted energy and increased TKE. The authors attributed this decrease not only to energy extraction, but also to the mixing induced by the increase in TKE. 11

13 As the research discussed above indicates, GCMs and mesoscale models do not allow for simulation of wind turbine wakes and their effect on boundary layer mixing. A recent study by Lu and Porté-Agel (2011) employed a large-eddy simulation (LES) for the purpose of presenting a more realistic approach to modeling wind turbines in a stable boundary layer (SBL). With the added computational expense of LES, a real case of a SBL was not feasible and an idealized case was utilized. The wind farm was parameterized with airfoil data and loading was determined by local angle of attack, blade movement, and local flow field. This allowed for the mixing mechanisms associated with blade motions to be represented and provide for increased realism of the wind turbine wakes. In addition, the actuator line model was used. This allowed for the simulation of helicoidal tip vortices and provided a more realistic turbulence structure. Two simulations were conducted with wind turbines to observe the effects of turbine spacing. The boundary layer height was observed to increase for both simulations resulting from the expansion of wind turbine wakes. The wind turbine simulation with a larger CD resulted in a larger boundary layer height increase. The larger CD wind-farm also resulted in a lower amount of power extraction and larger turbulence intensity, which agrees with the results of Adams and Keith (2011). Observational data from a wind-farm at San Gorgonio, California collected during June 18-August 9, 1989 were analyzed to determine wind-farm effects on surface air temperatures, model performance for near-surface air temperatures within wind farms, and mitigation of wind energy impacts (Baidya Roy and Traiteur 2010). Data indicated that wind turbine induced turbulence leads to enhanced mixing which acts to increase nocturnal surface temperatures and decrease daytime surface temperatures. The Regional Atmospheric Modeling System (RAMS) was employed to simulate a small wind farm, which was parameterized similar to methods of 12

14 Baidya Roy et al. (2004). The model was able to successfully model the surface temperature patterns that were observed at the San Gorgonio wind farm. When assessing impact mitigation, the authors observed that high TKE associated with large amounts of vertical wind shear decreased the significance of wind turbine induced turbulence and resulted in a much smaller change in surface temperature. The authors conducted a number of simulations with TKE added by rotors varying from low to high. It was found that as rotor TKE increased the amount of kinetic energy available for extraction by turbines further downwind was greatly reduced. However, it was found that a small amount of TKE is favorable as it acts to replace extracted energy by entraining momentum from flow outside of the wind farm. From these observations, the authors concluded that altering wind turbine design and wind farm siting are effective methods for mitigating impact. The Great Plains was cited as being a low impact site as turbulence is already large from high wind speeds and significant amounts of vertical wind shear. Data & Methodology Modeling NCAR s Weather Research and Forecast (WRF) model will be utilized for this research (Shamarock et al. 2008). The Advanced Research WRF (ARW) is an atmospheric model that is designed specifically for research applications. ARW is especially appealing because it is maintained as a community model, providing it with the advantages of input from a wide range of sources and allowing for it to quickly accommodate new physics schemes. ARW solves the fully compressible, Euler nonhydrostatic equations. It is capable of spatial resolutions suitable for mesoscale modeling all the way down to resolutions required for large eddy simulations (LES). For the purpose of modeling wind farms, there is currently a parameterization (Blahak et al. 2010) that is readily accessible via namelist options and another 13

15 parameterization expected to be included in the next ARW release (ARW v3.4) (Adams and Keith 2011). NCEP s North American Regional Reanalysis (NARR) will be used to initialize ARW and periodically provide data for boundary conditions (Shafran et al. 2006). NARR is composed of the NCEP/DOE Global Reanalysis, the NCEP Eta model, the Noah land surface model, and an abundance of additional observations. This reanalysis was developed for the purpose of providing a long-term data set with improved resolution and accuracy over that of the NCEP/DOE Global Reanalysis. The horizontal resolution is 32 km with 45 vertical layers. Data sets are available at 3-hour intervals, which allows for more frequent updating of boundary conditions. The suitability of WRF for simulation of LLJ events in the Great Plains was addressed by Storm et al. (2009) to determine its effectiveness in forecasting the LLJ. The purpose for their research was to improve forecasting of the LLJ for the sake of preventing turbine damage associated the large amount of shear stress induced by a strong LLJ event. Storm et al. (2009) determined that for all of the boundary layer schemes tested, WRF is capable of capturing some of the essential characteristics (e.g. location and timing) of the observed LLJ events (Storm et al. 2009). Model Setup WRF will be run with two nested grids at horizontal spatial resolution of 36 km for the outer grid and 12 km for the inner grid. The domains are shown in Figure 2. There will be 43 vertical levels from the surface to the top of the model with the highest vertical resolution being in the PBL. The lowest 140 meters will contain 7 vertical levels in order to provide the highest resolution around wind turbine hub height. Above 140 meters the vertical resolution will 14

16 decrease to 80 meters until 1 km where it will transition to 250 meters until the top of the model. PBL Schemes Two PBL schemes will be utilized in this research. The MYNN (Nakanishi and Niino 2009) scheme and the YSU (Hong et al. 2006) scheme. Both schemes have the benefit of being long-lived in the modeling community and have been subject to much improvement and extensive testing. MYNN is an improvement upon the MYJ level-2.5 model (Yamada 1977), which was developed from the original MYJ closure introduced by Mellor and Yamada (1974). MYNN made improvements by updating closure constants with LES data, formulating a new diagnostic equation for turbulent length scale, and improving stability functions. It is a 2.5 level local closure containing prognostic equations for mean variables and TKE. Reynolds stresses and second-order terms are determined diagnostically by neglecting time-tendency, advection and diffusion terms. This leaves dissipation rates, pressure covariances, and third-order moments to be parameterized. Dissipation rates are parameterized in terms of turbulent scale length, with the introduction of a diagnostic equation for turbulent scale length being one of the most significant improvements made to the scheme. Pressure covariance parameterization incorporates buoyancy effects, which were neglected in the MYJ scheme. Third order moments are parameterized as a function of turbulent length scale and stability functions. In testing the MYNN scheme against an LES simulation for a transitioning convective PBL, the authors observed that the scheme performed very closely to the LES simulation and was able to accurately capture evolution of the PBL. The YSU PBL scheme is a first order closure scheme with non-local mixing. This scheme evolved from the MRF non-local PBL scheme developed by Hong and Pan (1996). 15

17 Modifications were made to PBL schemes after Noh et al. (2003) based upon large eddy simulation (LES) data. Modifications made were the addition of explicit treatment of entrainment processes, the inclusion of vertically varying parameters such as Prandtl number in formulation of turbulent flux and eddy diffusivity, and inclusion of non-local mixing in eddy diffusivity of momentum. The turbulent diffusion equation for prognostic variables is given by (Noh et al. 2003): C t = + z K % C c z γ ( - ' c* w'c', & ) 3 % (. * 0 & h) / z ( ) h ' where C is a quantity such as momentum, heat or turbulence, K c is the eddy diffusivity, γ c is the z ( ) h % counter gradient term, and the entire term w'c' # & ( $ h' 3 represents the effects of entrainment at the top of the PBL and is the major addition made by Noh et al. (2003). The countergradient term provides a more realistic representation of turbulence in the PBL as quantities in the atmosphere are commonly transported by turbulent eddies. When compared against to the MYJ and ACM2, the YSU scheme performed closest to observations and significantly outperformed the MYJ scheme (Hu et al. 2010). Wind Farm Parameterizations Two wind farm parameterizations will be utilized in this research (Adams and Keith 2011; Blahak et al. 2010). Both parameterizations represent wind turbines as sub-grid scale features and are implemented as part of the boundary layer parameterization. The wind farm scheme introduced by Blahak et al. (2010), utilizes the power equation: ρ P(v rh ) = C l 0 p 2 v 3 π rh 4 d 2 r 16

18 where C p is the power coefficient, ρ l0 is a reference air density, v rh is wind velocity at rotor height, and d r is the diameter of the rotor. The time rate of atmospheric energy loss is derived from a discretized form of this equation containing a wind turbine density coefficient to represent the number of turbines per unit area. This scheme assumes C p is constant similar to methods used by Baidya-Roy and Pacala (2004). The amount of TKE added to the atmosphere is taken to be a flux of the total energy dissipation multiplied by a proportionality constant, which is limited by conservation of energy. This does not provide for a physically realistic representation of the behavior of turbulence, but currently seems to be a viable approach given the general lack of observations. A noteworthy limitation of this scheme is that it can only be used with the MYNN 2.5 level PBL scheme. The wind farm parameterization provided by Adams and Keith (2011) is considerably different from that of Blahak et al. (2010). The amount of energy removed from the atmosphere by wind turbines is represented by drag force as: F drag = 1 2 C T ρψv 2 where C t is the thrust coefficient, ρ is air density, ψ is the wind farm density, and V is the wind velocity. The thrust coefficient is defined as the force on the wind turbine from the pressure difference across the turbine and varies with wind speed (Burton et al. 2001). The amount of wind turbine induced TKE added to the atmosphere is parameterized by the same method Blahak et al. (2010) used to parameterized energy removal by wind turbines. The equation used here is: P TKE = 1 2 C TKEρψv 3 17

19 where C TKE = C T C p. This method allows for a more physically based parameterization of TKE addition by wind-turbines as the thrust and power coefficients vary with wind speed and represent wind-turbine performance. Defining the LLJ As previously discussed, there have been many definitions of a LLJ focusing on either vertical structure or horizontal structure depending on the nature of the research being conducted. This research will require a definition that addresses all three spatial dimensions as well as evolution through time. The criteria for low-level wind maxima to be classified as a NLLJ are: 1. Horizontal dimensions of the NLLJ will be defined as those areas having meridional components that are supergeostrophic by at least 50 percent. In addition, regions must have a meridional velocity greater than or equal to 10 m/s. 2. From the surface to the height of the NLLJ maximum, wind speeds must transition from subgeostrophic to supergeostrophic. Above the height of the NLLJ maximum, the meridional velocity must decrease to subgeostrophic speeds before reaching the PBL top. If the PBL top is higher than 3 km AGL, the NLLJ must decrease to subgeostrophic speeds before reaching 3 km AGL. The PBL height will be defined as the level at which the lapse rate transitions from being stable to neutral or unstable. This criteria ensures that vertical structure of the meridional wind has a jet-like structure. 3. With the focus of this study being on nocturnal LLJs, there must be an observed diurnal evolution in the vertical structure of the meridional wind. In addition to requiring an increasing magnitude of the meridional wind throughout the transition from convective to stable PBL, the wind profile maximum for the most unstable part of the day must not meet the criteria of Part 2. 18

20 Case Studies As a result of the complex combination of forcing mechanisms that may be present in any single NLLJ event, it is necessary to examine an ensemble of events in an effort to determine the long-term impacts of wind energy on the NLLJ. Changes to the NLLJ would be most significant during the warm season because it contains the strongest NLLJs and a decrease in moisture flux would have the greatest societal impact. Simulations will be run for May, June, and July of This will provide a large enough set of NLLJ events to give the results of the study statistical significance and to provide a realistic estimate of long-term potential impacts. Wind-Farm Location, Size, and Capacity Density For all cases, the wind-farm will be centered at the climatological maximum of the NLLJ (Figure 3). This maximum was determined by Rife et al. (2010) by analysis of 21 years of high resolution model reanalysis data. The NLLJ maximum coincides well with the maximum in the National Renewable Energy Laboratory (NREL) 80 meter wind energy resource providing for realistic wind-farm siting. A study by NREL found that CDs for 161 land-based, US wind energy projects ranged from 1.0 to 11.2 MW/km 2. The average CD for these projects was found to be 3.0 MW/km 2 with the vast majority of the projects falling between 1.0 and 6.0 MW/km 2 (Denholm et al. 2009). Capacity densities of 3.0 W/m 2 and 6.0 W/m 2 will be utilized in order to simulate the average and practical upper limit of wind-farm density. Wind farm sizes will vary by running simulations for three different cases of wind energy growth: 1. Wind Force 12 (Millais and Teske 2003) This is a blueprint for having 12% of the world s electricity come from wind energy by U.S. installed wind energy capacity will need to be 250 GW to meet his goal. 19

21 2. 20% Wind Energy by 2030 (Salerno et al. 2008) This scenario requires that 20% of all U.S. electricity generation comes from wind energy by Based on AWEA estimates, this would require installed capacity to be around 300 GW. 3. Stabilization Wedge from Pacala and Socolow (2004) In order for wind energy to serve as a stabilization wedge there would need to 2000 GW of worldwide installed capacity added by As the focus here is on one region of the U.S., the U.S. portion of the worldwide wind energy growth used in the Wind Force 12 blueprint will also be used here (~400 GW). This amounts to 417 GW of installed capacity, which will be rounded down to 400 GW for simplicity. To put these goals into perspective, the US Department of Energy reported that US installed capacity was 46.9 GW as of December To assess whether NLLJ response is sensitive to its location with respect to the wind-farm, individual case studies will be analyzed for NLLJs located: near the center of the wind-farm and upwind, downwind, and east of the wind-farm. The jet location will be determined by taking the average over 0200 UTC to 1300 UTC. Analyzing a NLLJ east of the wind-farm will provide for mixing down of warmer temperatures over sloping terrain and allow for a case of mixing out baroclinicity. Methods of Analysis There are a number of variables that will serve as useful tools in not just determining the degree to which wind-farms affect the NLLJ, but also the specific process that are producing the observed response. Being a study that is focused on a process occurring exclusively in the stable boundary layer, many of the variables utilized will be unique to boundary layer meteorology. The following variables have potential for assisting in analysis, but some have not yet been tested for use in this study: 20

22 1. Turbulent Kinetic Energy (TKE) will allow for a direct observation of the amount of turbulence that is being added to the atmosphere and will assist in diagnosing impacts to baroclinicity and eddy viscosity. 2. Static Stability (θ) will be useful for diagnosing the degree to which TKE is mixing the PBL. In particular, this variable will be useful for determining how the redistribution of energy by enhanced mixing is effecting baroclinicity. 3. Bulk Richardson Number (R B ) is a non-dimensional number that compares buoyancy and shear to determine dynamic stability. The American Meteorological Society (AMS) gives R B as: R B = (g /T v )Δθ v Δz (ΔU) 2 + (ΔV ) 2, where g is gravity, T v is absolute virtual temperature, θ v is virtual potential temperature, Δz is thickness, and ΔU and ΔV are changes in horizontal velocity across that thickness. All differences will be taken by central difference methods using adjacent vertical levels. This variable will be useful in determining the degree to which wind-farm induced TKE changes shear induced turbulence. 4. Thermal Wind is the vertical shear of the geostrophic wind and will be useful in determining the degree to which baroclinicity is changed in the presence of enhanced mixing. Thermal wind will not be directly calculated as its strength and direction can be inferred from temperature gradient. 5. Additional variables will be examined as needed. Preliminary Results A control simulation and a wind farm simulation were run for June The wind farm was placed at the center of the mean NLLJ maximum. It extended over an area of 56,160 km 2 with a CD of 5 W/m 2 and total installed capacity of GW with each turbine being 2 MW. 21

23 This installed capacity for this case was between the 20% Wind Energy by 2030 and Wind Force 12 scenarios. Seventeen of the nights were classified as having a NLLJ. The central portion of the mean NLLJ was located over central Kansas with a maximum average meridional velocity of 12 m/s. Figure 4 shows the average NLLJ meridional velocity at the height of maximum velocity. The primary NLLJ location throughout June 2002 was from northern Oklahoma extending through central Kansas and into southern Nebraska. The definition of geostrophic wind was taken from a spatial and temporal average of the study area from 0200 UTC to 12 UTC (sunset to 1 hr after sunrise). Rather than use a spatial average of geostrophic wind for classification, future work will classify each grid point according to the geostrophic wind values at that grid point. All vertical levels meeting the criteria for a NLLJ will then be summed to provide the total NLLJ for that event. After establishing the horizontal dimensions of the NLLJ, vertical profiles of meridional velocity will be constructed to further determine if the event meets the NLLJ criteria. This will be discussed in further detail later. Figure 5 shows the difference between the wind farm simulation and the control simulation. The primary decrease in NLLJ magnitude occurred at the wind farm with a maximum decrease of 5 m/s and extended downwind of the wind farm by almost 250 km. There were two locations to the east of the wind farm that responded unexpectedly. Directly to the east of the wind farm there was an increase in NLLJ magnitude by as much as 3 m/s. A possible reason for this increase is a slight shift in NLLJ location, but diagnosing the cause of this response would require some additional investigation. The second region to respond unexpectedly was located in central Oklahoma. The NLLJ in this region decreased by as much as 3 m/s despite the fact that it was located almost 200 km to the south of the wind farm. 22

24 Looking back at Figure 4 it can be seen that this region was a predominant location for NLLJ formation. It is possible that this response was simply a shift in NLLJ location, but being so far upwind of the wind farm suggests that a case study of a wind farm downwind of a NLLJ should provide some interesting results. To provide examples of how NLLJ classification and some of the methods of analysis will be accomplished, the following section will now focus on a NLLJ that occurred on June 2, Figure 6 shows the horizontal dimensions of this NLLJ. It was also located approximately within the wind farm and reached a maximum temporally averaged meridional velocity of 22 m/s at 400 meters AGL. Figure 7 shows the difference between wind farm the simulation and the control simulation. Similar to the feedbacks experienced for the entire month, the June 2 nd NLLJ experienced a reduction in velocity in and downwind of the wind farm. The northern most decrease was in excess of 10 m/s and can be interpreted as a shift in the NLLJ lateral boundary. Being located downwind of the wind farm, this decrease does follow what would be expected. The rest of the NLLJ experienced a slight growth in its lateral extent, which can be seen by the higher velocities located along its previously defined lateral extent. It is unclear why the NLLJ expanded in these directions, but the response was so insignificant that it could be the result of minor changes in the horizontal pressure gradient or some other forcing mechanism. Figure 8 shows vertical profiles of meridional velocity throughout NLLJ growth for the control and wind farm simulations. The profiles were constructed by taking spatial averages over the regions shown in Figure 7 that experienced a decrease in velocity. The method for classifying the NLLJ is demonstrated here. The boundary layer achieved its lowest static stability at 2000 UTC on June 1 st making it the reference point for determining if the LLJ is nocturnal. It can be seen that the wind profile was subgeostrophic at this time; it did transition to 23

25 supergeostrophic above 1.7 km MSL, but did not return to subgeostrophic before the top of the PBL, which satisfies the 3 rd criteria for NLLJ classification. The maximum velocity occurred at 0600 UTC reaching a velocity of m/s at 0.9 km. It can be seen that from the surface and prior to 0.9 km the meridional velocity was subgeostrophic; it then transitioned to supergeostrophic and remained so until 1.5 km where it transitioned back to being subgeostrophic, which meets the 2 nd criteria for NLLJ classification. From 2000 UTC to 0600 UTC there was a constant increase in meridional velocity throughout the entire depth of the boundary layer as the PBL transitioned from turbulent to laminar flow. The wind farm simulation followed the same pattern of evolution as the control simulation, but with consistently lower velocities. Differences became larger throughout NLLJ growth with differences reaching less than 1 m/s during daytime and 2 m/s at the time of the maximum. The height of the maximum was consistently higher for the wind farm simulation than it was for the control. This is most likely a result of TKE increasing PBL depth as was observed by Lu and Porté-Agel (2011). Figure 9 shows NLLJ decay from 0600 UTC to 14 UTC. After 0600 UTC the difference between the control simulation and the wind farm simulation became much more pronounced, reaching differences as high as 2.5 m/s. A notable feature was the rebounding of the NLLJ from 1000 UTC to 1200 UTC. Figures 10 and 11 partially explain this feature as being forced by baroclinicity. In Figure 10 we have meridional velocity and horizontal temperature gradient for the control simulation. The horizontal temperature gradient serves as a surrogate for thermal wind with the values shown indicating a southerly component of thermal wind. In Figure 10 the velocity increased at 99.2W longitude at 1200 UTC, which coincided with an increase in thermal wind. In Figure 11 we see that the for the wind farm simulation the increase at 1200 UTC was 24

26 almost undetectable and coincided with a much weaker thermal wind. Temperature gradient distribution was also much different between simulations and appears to be correlated to the differences in velocity distribution. Figure 12 shows vertical profiles of TKE from 0200 UTC to 1200 UTC. As expected, the wind farm induced a net increase in TKE. At 0200 UTC the TKE difference between the control and wind farm simulations was the largest as the NLLJ had not yet evolved and shear generated turbulence was at a minimum. Progressing through NLLJ evolution the TKE difference between the two simulations converged with the control simulation TKE being larger than the wind farm simulation TKE at 0600 UTC and 0.5 km, which coincided with the time and height of the largest amount of shear generated turbulence. TKE increased at 1200 UTC with transition from a stable to convective PBL. At this time the control simulation TKE was larger than the wind-farm simulation TKE until ~650 meters, once again because of shear generated turbulence. These results support Baidya Roy s (2010) designation of the Great Plains as a low impact region with respect to turbulence associated with wind energy applications. Figure 13 shows evolution of static stability profiles from 2000 UTC to 0600 UTC. The unstable boundary layer profiles (2000 and 0000 UTC) were cooler and more unstable for the wind farm simulation throughout most of the PBL. Stable boundary layer profiles ( UTC) were warmer and more stable until ~2 km AGL. It was expected that the wind farm would induce surface cooling during the day and surface warming at night, but it was not expected that the nocturnal PBL would be more stable than the wind-farm simulation. Looking back at Figure 12 we can see that above 0.7 km MSL TKE was much larger for the wind-farm simulations at all times. In addition to being larger, the difference in TKE between the two simulations increased with height indicating a larger amount of mixing at higher levels. A greater amount of mixing 25

27 would create more warming and therefore set up a more stable lapse rate much like the static stability profiles observed in Figure 13. Preliminary Conclusions The preliminary results indicate that atmospheric feedbacks induced by wind-farms are non-negligible for NLLJ formation. Despite being located in the lowest 200 meters of the PBL, wind-turbines induce changes throughout the entire depth of the PBL through extraction of energy and addition of turbulence. Furthermore, the atmosphere s response to these feedbacks is classified by many complex interactions, especially when looking at a feature such as the NLLJ. The degree to which wind-farm induced TKE will change the NLLJ is a function of; NLLJ strength, height, vertical and horizontal distribution, wind farm size, density, and location and many other factors that play a significant role in the magnitude of these induced changes. References Adams, A. S., and D. W. Keith, 2011: Are global wind power resource estimates overstated?, submitted. Archer, C. L., and M. Z. Jacobson, 2005: Evaluation of global wind power. Journal of Geophysical Research, 110, Baidya Roy, S., S. W. Pacala, and R. L. Walko, 2004: Can large wind farms affect local meteorology? J. Geophys. Res., 109, D Baidya Roy, S., and J. J. Traiteur, 2010: Impacts of wind farms on surface air temperatures. PNAS, 107, Barrie, D. B., and D. B. Kirk-Davidoff, 2010: Weather response to a large wind turbine array. Atmos. Chem. Phys., 10, Blackadar, A. K., 1957: Boundary Layer Wind Maxima and Their Significance for the Growth of Nocturnal Inversions. Bulletin of the American Meteorological Society, 38, Blahak, U., B. Goretzki, and J. Meis, 2010: A simple parameterization of drag forces induced by large wind farms for numerical weather prediction models. Proceedings, European Wind Energy Conference,

28 Bonner, W. D., 1968: Climatology of the low-level jet. Mon. Wea. Rev., 96, Buajitti, K., and A. K. Blackadar, 1957: Theoretical studies of diurnal wind structure in the planetary boundary layer. Quart. J. Roy. Meteor. Soc., 83, Burton, T., D. Sharpe, N. Jenkins, and E. Bossanyi, 2001: Wind Energy Handbook. John Wiley & Sons, Ltd. Chen, Y. L., X. A. Chen, and Y. X. Zhang, 1994: Diagnostic study of the low level jet during TAMEX IOP 5. Mon. Wea. Rev., 122, Denholm, P., M. Hand, M. Jackson, and S. Ong, 2009: Land-Use Requirements of Modern Wind Power Plants in the United States. NREL Tech. Rep. NREL/TP-6A , 39 pp. Fawbush, W. J., R. C. Miller, and L. G. Starrett, 1951: An empirical method of forecasting tornado development. Bull. Amer. Meteor. Soc., 32, 1-9. Fraquharson, S. J., 1939: The diurnal variations of wind over tropical Africa. Quart. J. Roy. Meteor. Soc., 65. Goualt, J., 1938: Vents en altitude a fort Lamy (Tchad). Ann. Phys. du Globe de la France d'outre-mer, 5, Helfand, H. M., and S. D. Schubert, 1995: Climatology of the simulated Great Plains low-level jet and its contribution to the continental moisture budget of the United States. J. Climate, 8, Higgins, R. W., Y. Yao, E. S. Yarosh, J. E. Janowiak, and K. C. Mo, 1997: Influence of the Great Plains low-level jet on summertime precipitation and moisture transport over the central United States. J. Climate, 10. Hoffert, M. I., and Coauthors, 2002: Advanced Technology Paths to Global Climate Stability: Energy for a Greenhouse Planet. Science, 298, Holton, J. R., 1967: The diurnal boundary layer wind oscillation above sloping terrain. Tellus, 19, Hong, S.-Y., and H. -L. Pan, 1996: Nonlocal boundary layer vertical diffusion in a Medium- Range Forecast model. Mon. Wea. Rev., 124, Hong, S.-Y., Y. Noh, and J. Dudhia, 2006: A New Vertical Diffusion Package with and Explicit Treatment of Entrainment Processes. Mon. Wea. Rev., 134, Hu, X. M., Nielson-Gammon, J. W., and F. Zhang, 2010: Evaluation of three planetary boundary layer schemes in the WRF model. J. Appl. Meteor. and Climatol, 49, Jiang, X., N.-C. Lau, I. M. Held, and J. J. Ploshay, 2007: Mechanisms of the Great Plains Low- Level Jet as Simulated in an AGCM. Journal of the Atmospheric Sciences, 64, Keith, D. W., J. F. DeCarolis, D. C. Denkenberger, D. H. Lenschow, S. L. Malyshev, S. Pacala, and P. J. Rasch, 2004: The influence of large-scale wind power on global climate. 27

29 Proceedings of the National Academy of Sciences of the United States of America, 101, Kirk-Davidoff, D. B., and D. W. Keith, 2008: On the climate impact of surface roughness anomalies. Journal of Atmospheric Sciences, 65, Kis, A. K., and J. M. Straka, 2009: Nocturnal Tornado Climatology. Wea. and Forecasting, 25, Kraus, H., J. Malcher and E. Schaller, 1985: Nocturnal low-level jet during PUKK. Bound. - Layer Meteor., 31, Lu, H., and F. Porté-Agel, 2011: Large-eddy simulation of a very large wind farm in a stable atmospheric boundary layer. Phys. Fluids, 23, Means, L. L., 1952: On thunderstorm forecasting in the central United States. Mon. Wea. Rev., 80, , 1954: A study of the mean southerly wind-maximum in low levels associated with a period of summer precipitation in the middle west. Bull. Amer. Meteor. Soc., 35, Mellor, G. L., and T. Yamada, 1974: A Hierarchy of Turbulence Closure Models for Planetary Boundary Layers. J. Atmos. Sci., 31, Millais C., and S. Teske, 2003: Wind Force 12 - A Blueprint to Achieve 12% of the World's Electricity from Wind Power by EWEA/Greenpeace. Modahl, A. C., 1979: Low-level wind and moisture variations preceding and following hailstorms in northeast Colorado. Mon. Wea. Rev., 107, Nakanishi, H., and H. Niino, 2009: Development of an improved turbulence closure model for the atmospheric boundary layer. J. of the Met. Soc. of Jap., 87, Noh, Y., W. G. Cheon, S. -Y. Hong, and S. Raasch, 2003: Improvement of the K-profile model for the planetary boundary layer based on large eddy simulation data. Bound. -Layer Meteor., 107, Pacala, S., and R. Socolow, 2004: Stabilization Wedges: Solving the Climate Problem for the Next 50 Years with Current Technologies. Science Magazine, 305, Pielke, J., R., 2010: The climate fix: what scientists and politicians won't tell you about global warming. Basic Books. Rasmussen, E. M., and D. O. Blanchard, 1998: A Baseline Climatology of Sounding-Derived Supercell and Tornado Forecast Parameters. Wea. and Forecasting, 13, Reiter, E. R., 1963: Jet Stream Meteorology. Universtiy of Chicago Press, 515 pp., 1969: Tropopause circulation and jet streams. Climate of the Free Atmosphere. World Survey of Climatology, E. D. F. Rex, Ed., Elsevier,

30 Rife, D. L., J. O. Pinto, A. J. Monaghan, C. A. Davis, and J. R. Hannan, 2010: Global Distribution and Characteristics of Diurnally Varying Low-Level Jets. J. of Climate, 23, Salerno, E., R. Gramlich, A. Silverstein, and P. Donnelly, 2008: 20% Wind Energy by Increasing Wind Energy's Contribution to U.S. Electricity Supply. U. S. D. o. Energy, Ed., 1. Shafran, P. and Coauthors, 2006: North american regional reanalysis. Bull. Amer. Meteor. Soc., 87, Shapiro, A., and E. Fedorovich, 2009: Nocturnal low-level jet over a shallow slope. Acta Geophysica, 57, , 2010: Analytical description of a nocturnal low-level jet. Quarterly Journal of the Royal Meteorological Society, 136, Skamarock, W. C., J. B. Klemp, J. Dudhia, D. O. Gill, D. M. Barker, M. G. Duda, X. Huang, and W. Wang, J. G. Powers, 2008: A description of the Advanced Research WRF Version 3. NCAR Tech Note-475+STR. Song, J., K. Liao, R. L. Coulter, and B. M. Lesht, 2005: Climatology of the Low-Level Jet at the Southern Great Plains Atmospheric Boundary Layer Experiment Site. Journal of Applied Meteorology, 44, Storm, B., and S. Basu, 2010: The WRF Model Forecast-Derived Low-Level Wind Shear Climatology over the United States Great Plains. Energies, 3, Storm, B., J. Dudhia, S. Basu, A. Swift, and I. Giammanco, 2009: Evaluation of the Weather Research and Forecasting model on forecasting low-level jets: implications for wind energy. Wind Energy, 12, Stull, R. B., 1988: An Introduction to Boundary Layer Meteorology. Vol. 13, Springer. Tollerud E. I., and Coauthors, 2008: Mesoscale moisture transport by the low-level jet during the IHOP field experiment. Mon. Wea. Rev., 136, Uccellini, L. W., 1980: On the role of upper tropospheric jet streaks and leeside cyclogenesis in the development of low-level jets in the Great Plains. Mon. Wea. Rev., 108, Uccellini, L. W., and D. R. Johnson, 1979: The coupling of upper and lower tropospheric jet streaks and implications for the development of severe convective storms. Mon. Wea. Rev., 107, Wagner, A., 1939: Uber die Tageswinde in der freien Atmosphare. Beitr. Phys. Atmos., 25. Wang, S.-Y., and T.-C. Chen, 2009: The Late-Spring Maximum of Rainfall over the U.S. Central Plains and the Role of the Low-Level Jet. Journal of Climate, 22, Werth, D., R. Kurzeja, N. L. Dias, G. Zhang, H. Duarte, M. Fischer, M. Parker, and M. Leclerc, 2011: The Simulation of the Southern Great Plains Nocturnal Boundary Layer and the 29

31 Low-Level Jet with a High-Resolution Mesoscale Atmospheric Model. J. Appl. Meteor. and Climatol., 50, Whiteman, C. D., X. Bian, and S. Zhong, 1997: Low-level jet climatology from enhanced rawinsonde observations at a site in the southern Great Plains. J. Appl. Meteor., 36,

32 Land-Based Wind Resource United States - Annual Average Wind Speed at 80 m Wind Speed m/s > < 4.0 Source: Wind resource estimates developed by AWS Truepower, LLC for windnavigator. Web: Spatial resolution of wind resource data: 2.5 km. Projection: Albers Equal Area WGS APR Fig. 1. Figure was provided by the National Renewable Energy Laboratory. Annual average wind speed at 80 meters (hub-height used in this study). The Great Plains region is by far the most significant resource with wind speeds commonly reaching 8 to 8.5 m/s. 31

33 Model Domains Fig. 2. The outer domain (resolution of 36 km) extends from coast to coast and from Panama to the Arctic Circle. The inner domain (resolution of 12 km) covers the entire region of interest. 32

34 United States - Annual Average Wind Speed at 8 Wind-Farm Location United States - Annual Average Wind Speed at 80 m b" a" Center of Wind-Farm at 36.4 N, 100 W Wind Speed m/s > < 4.0 Source: Wind resource estimates developed by AWS Truepower,. Web: LLC windnavigator Fig. 3. The wind-farm for all simulations will be centered atfor36.4 N, 100 W. This location was chosen Spatial resolution of wind resource because it is (a) the climatological maximum of the NLLJdata: from Rife et al. (2010) and (b) lies along the 2.5 km. Projection: Albers Equal Area WGS84. maximum annual average 80 meter wind speed (hub-height used in this study). Source: Wind resource LLC for windnavigator data: 2.5 km. Projection 01-APR

35 June 2002 Control NLLJ Wind" Farm" m/s" Fig. 4. The mean NLLJ for June There were a total of 17 individual NLLJ events. The most predominant location was in central KS with an average of 12 m/s. 34

36 June 2002 With Wind Farm minus Without Wind" Farm" m/s" Fig. 5. The wind farm simulation NLLJ minus the control simulation NLLJ. The largest decrease in meridional velocity occurred in and downwind of the windfarm with differences as large as 5 m/s. 35

37 Nocturnal Low-Level Jet (June 2) Wind" Farm" m/s" Fig. 6. Meridional velocity component averaged from 0200 UTC to 1300 UTC. Average was taken at the height of the NLLJ maximum (~400 m AGL). The average reached a maximum of 22 m/s. 36

38 June 2, 2002 With Wind Farm minus Without Wind" Farm" m/s" Fig. 7. June 2 nd wind-farm simulation minus the control simulation. Similar to the entire month, the larges decrease was in and downwind of the wind-farm. Shifts in lateral boundary are especially evident along the edges of the NLLJ. 37

39 1.5 Low Level Jet Growth control wind farm Height MSL (km) z Vg 20z 04z 06z z 22z 00z Vg 06z V (m/s) Fig. 8. Vertical profiles of the meridional wind spatially averaged from all areas experiencing a decrease in velocity as seen in Fig. 7. There is a set of profiles for the wind farm simulation and a set for the control simulation showing the growth of the NLLJ from 2000 UTC to 0600 UTC. Geostrophic wind profiles for 2000 UTC and 0600 UTC are also plotted for use in jet classification. Evolution from a well-mixed boundary layer to a stable boundary layer with a well-defined NLLJ is evident for both simulations. 38

40 1.5 Low Level Jet Decay control wind farm z Height MSL (km) z 12z 10z 12z 06z V (m/s) Fig. 9. Same as Fig. 8. except decay of the NLLJ is shown from 0600 UTC to 1400 UTC. The difference between the wind-farm simulation and the control simulation was much more pronounced throughout decay. The most interesting feature was the increase in NLLJ magnitude that occurred from 1000 UTC to 1200 UTC. This is believed to be forced in part by baroclinicity. 39

41 June 2, 2002 Control Simulation Baroclinicity Meridional Velocity Temperature Gradient Increase Increase m/s K/100km Fig. 10. Panel A is a Hovmöller plot of meridional velocity taken at a constant latitude through the maximum velocity decrease at 1 km. Panel B is a Hovmöller plot of east-west temperature gradient taken at the same latitude and at 0.7 km. The east west temperature gradient serves as a surrogate for thermal wind. The y-axis ranges from 0000 UTC to 1500 UTC. The most important feature is the increase in velocity that occurred at 1200 UTC around 99W longitude. The right panel shows a maximum in temperature gradient occurring at the same time and location indicating some correlation between the two features. 40

42 June 2, 2002 Wind-Farm Simulation Baroclinicity Meridional Velocity Temperature Gradient Increase Increase m/s K/100km Fig. 11. Same as Fig. 9. for the wind-farm simulation. The two features discussed are still present, but both are much weaker providing support for NLLJ response to wind-farms through changes in baroclinicity. 41